ETRI Launches No-Code Machine Learning Tools to Boost AI Access

The Electronics and Telecommunications Research Institute (ETRI) has unveiled a revolutionary no-code machine learning development framework designed to simplify AI deployment for users with limited technical expertise. Since 2021, ETRI has focused on enhancing accessibility to artificial intelligence in various sectors, including manufacturing, healthcare, and shipbuilding. On November 6, 2023, the institute announced the release of its MLOps tool as open source on GitHub, marking a significant step in democratizing AI technology.

Details of the TANGO Framework

The TANGO framework allows users to automatically generate neural networks without the need for intricate coding skills. This no-code approach not only streamlines the deployment process but also makes it easy for individuals in industrial fields to utilize AI effectively. For instance, while identifying defects in steel during quality inspections is straightforward, integrating AI solutions has historically posed challenges. Similarly, doctors can diagnose tuberculosis from X-ray images, yet developing AI-driven predictive models has been complex.

ETRI’s TANGO framework addresses these challenges by offering a user-friendly interface that enables rapid application software development and deployment across diverse hardware environments, including cloud and on-device systems. Users can install the framework with a simple command and immediately access its functionalities via a web interface.

In traditional AI development, domain experts typically managed data labeling, while software developers focused on model training and application execution. The growing demand for software solutions across industries has highlighted a significant shortage of AI and software specialists. To combat this, ETRI has optimized its neural network automation algorithm for object recognition, tailoring it to meet the specific needs of domestic industries.

Commercialization and Future Plans

The launch of TANGO has also facilitated the development of LLMOps tools, further supporting generative AI applications. ETRI’s achievements include the filing of 24 domestic and international patents and multiple publications in prominent journals, contributing to approximately KRW 10 billion in commercialization revenue. One notable success story involves Avenotics, a company specializing in autonomous maritime navigation solutions. It secured an investment of 1.3 billion won through the “Public Research Outcome Expansion and Commercialization Project” backed by the Ministry of Science and ICT.

Through technology transfer, Avenotics has utilized TANGO’s on-device deployment technology and AI optimization capabilities to commercialize an AI system that generates contextual information for navigators. ETRI’s research team is actively promoting broader adoption of the TANGO framework, conducting pilot demonstrations with partner institutions.

Collaborative efforts have also emerged, such as a project with Weda Co., Ltd., which developed an AI service for visual inspections in the automotive parts manufacturing sector. This service targets intricate shapes, such as automotive bumper designs. Additionally, Lablup Inc. has partnered with KT Cloud to optimize deployment services for AI engines.

In healthcare, Seoul National University Hospital is leveraging AI technology to analyze chest CT scans and generate diagnostic reports automatically. This initiative aims to provide predictive services for cardiopulmonary diseases, with evaluations planned across four hospitals.

ETRI’s commitment to advancing AI development tools is evident through its plans for ongoing updates to the TANGO source code, with new versions set to be released every six months. The institute intends to hold annual public seminars, which have already attracted 944 participants from 552 institutions, fostering a rich exchange of insights regarding AI technology.

As noted by Kim Tae-ho, Software Project Manager at the Institute of Information & Communications Technology Planning & Evaluation (IITP), TANGO represents a pivotal open-source project in Korea, enhancing the nation’s competitiveness in AI development tools. Jo Chang Sik, Principal Researcher at ETRI, emphasized the importance of expanding TANGO into the realm of LLMOps, ensuring that the framework continues to evolve and support industry needs.

With these initiatives, ETRI is poised to drive significant advancements in machine learning accessibility, equipping a broader range of users with the tools needed to harness the potential of artificial intelligence.